[PATCH 0/9] Experimental S3C2410A cpufreq driver

[Werner Almesberger]

Cesar Eduardo Barros wrote:
> No, I didn't try to measure anything so far. In fact, I'm also wondering 
> how much of a difference it will make.

Good, so this ties in with our general hunt for the last lost
microwatt :-)

> My plan is to use parts of what has to be done to make suspend work and 
> combine it with cpufreq so the CPU can be always running but in a very 
> low clock speed.

Hmm, that seems to be an unusual scenario. In what cases do you think
this would be preferable to just doing a regular suspend ?

One potential scenario I was thinking of would be a low-overhead task
that doesn't need a lot of peripherals, e.g., MP3 playback. There, it
may be possible to get better overall power efficiency by running at a
constant low speed than being in stop-and-go mode with bursts of high
speed. (What would also make MP3 playback a good case is that we have
almost no buffering in hardware :-( )

- Werner